[20190505]关于latch 一些统计信息.txt
[20190505]关于latch 一些统计信息.txt
--//我在两篇文章,提到一些latch的统计信息.链接如下:
http://blog.itpub.net/267265/viewspace-2642329/ =>[20190423]简单测试latch nowilling等待模式.txt
http://blog.itpub.net/267265/viewspace-2641549/ =>[20190416]查看shared latch gets的变化.txt
--//我提到过以前我对这些统计是非常混乱的,到底什么情况是immediate_gets,什么情况下SPIN_GETS,sleep会增加.
--//我还是通过测试说明,一个测试胜过一大堆文字说明...
1.环境:
SCOTT@book> @ ver1
PORT_STRING VERSION BANNER
------------------------------ -------------- --------------------------------------------------------------------------------
x86_64/Linux 2.4.xx 11.2.0.4.0 Oracle Database 11g Enterprise Edition Release 11.2.0.4.0 - 64bit Production
>select addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where lower(name) like '%'||lower('test excl. parent l0')||'%'
ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
00000000600098D8 test excl. parent l0 0 4 151 0 0 0 0 0 0 0 0
--//我在链接http://blog.itpub.net/267265/viewspace-2642329/提到nowilling等待模式的情况,IMMEDIATE_GETS和IMMEDIATE_GETS的变化.不再论述.
--//如下测试仅仅关注SPIN_GETS,MISSES,SLEEPS的变化.
2.测试:
$ cat aa.sql
select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='test excl. parent l0';
host sleep 1
--//注意是test excl. parent l0 后面的不是10是l0.一直以为是10。看来要一副好眼力.
$ cat y1.sh
#! /bin/bash
zdate=$(date '+%Y%m%d%H%M%S')
echo $zdate
source peek.sh 'test excl. parent l0' 20 | timestamp.pl > /tmp/peekx_${zdate}.txt &
seq 20 | xargs -I{} echo -e 'sqlplus -s -l / as sysdba <<< @latch_free\nsleep 1' | bash > /tmp/latch_free_${zdate}.txt &
sqlplus -s -l / as sysdba <<EOF > /tmp/latch_sum_${zdate}.txt &
$(seq 20 | xargs -I {} echo @aa)
EOF
# 参数如下: @ exclusive_latch.txt latch_name willing why where sleep_num
sqlplus / as sysdba <<< "@ exclusive_latch.txt 'test excl. parent l0' 1 1 2 6" > /dev/null &
sleep 2
sqlplus / as sysdba <<< "@ exclusive_latch.txt 'test excl. parent l0' 1 3 4 6" > /dev/null &
sleep 10.2
sqlplus / as sysdba <<< "@ exclusive_latch.txt 'test excl. parent l0' 1 5 6 6" > /dev/null &
wait
$ cat exclusive_latch.txt
/* 参数如下: @ exclusive_latch.txt latch_name willing why where sleep_num */
--//connect / as sysdba
col laddr new_value laddr
SELECT addr laddr FROM v$latch_parent WHERE NAME='&&1';
oradebug setmypid
oradebug call kslgetl 0x&laddr &&2 &&3 &&4
host sleep &&5
oradebug call kslfre 0x&laddr
--//exit
3.执行:
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='test excl. parent l0';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- -------------------- ------ ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-04-30 12:13:39 00000000600098D8 test excl. parent l0 0 4 17 12 12 0 0 0 0 0 95671068
$ . y1.sh
20190430121417
[4]- Done sqlplus / as sysdba <<< "@ exclusive_latch.txt 'test excl. parent l0' 1 1 2 6" > /dev/null
[5]+ Done sqlplus / as sysdba <<< "@ exclusive_latch.txt 'test excl. parent l0' 1 3 4 6" > /dev/null
[1] Done source peek.sh 'test excl. parent l0' 20 | timestamp.pl > /tmp/peekx_${zdate}.txt
[4]+ Done sqlplus / as sysdba <<< "@ exclusive_latch.txt 'test excl. parent l0' 1 5 6 6" > /dev/null
[2]- Done seq 20 | xargs -I{} echo -e 'sqlplus -s -l / as sysdba <<< @latch_free\nsleep 1' | bash > /tmp/latch_free_${zdate}.txt
[3]+ Done sqlplus -s -l / as sysdba > /tmp/latch_sum_${zdate}.txt <<EOF
$(seq 20 | xargs -I {} echo @aa)
EOF
$ grep "00000000600098D8" /tmp/latch_sum_20190430121417.txt
--//行头是我添加上的。
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- -------------------- ------ ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-04-30 12:14:18 00000000600098D8 test excl. parent l0 0 4 18 12 12 0 0 0 0 0 95671068
2019-04-30 12:14:19 00000000600098D8 test excl. parent l0 0 4 18 12 12 0 0 0 0 0 95671068
2019-04-30 12:14:20 00000000600098D8 test excl. parent l0 0 4 18 12 12 0 0 0 0 0 95671068
2019-04-30 12:14:21 00000000600098D8 test excl. parent l0 0 4 18 12 12 0 0 0 0 0 95671068
2019-04-30 12:14:22 00000000600098D8 test excl. parent l0 0 4 18 12 12 0 0 0 0 0 95671068
2019-04-30 12:14:23 00000000600098D8 test excl. parent l0 0 4 18 12 12 0 0 0 0 0 95671068
--//过了6秒。gets数量增加1。MISSES,SLEEPS数量不变。
2019-04-30 12:14:24 00000000600098D8 test excl. parent l0 0 4 19 13 13 0 0 0 0 0 99676819
--//7秒后,第2个session请求获得latch（实际上等了4秒），统计更新gets增加1(也就是在获得latch时统计信息才更新).因为这次没有马上获取成功。
--//MISSES,SLEEPS数量增加1。
2019-04-30 12:14:25 00000000600098D8 test excl. parent l0 0 4 19 13 13 0 0 0 0 0 99676819
2019-04-30 12:14:26 00000000600098D8 test excl. parent l0 0 4 19 13 13 0 0 0 0 0 99676819
2019-04-30 12:14:27 00000000600098D8 test excl. parent l0 0 4 19 13 13 0 0 0 0 0 99676819
2019-04-30 12:14:28 00000000600098D8 test excl. parent l0 0 4 19 13 13 0 0 0 0 0 99676819
2019-04-30 12:14:29 00000000600098D8 test excl. parent l0 0 4 19 13 13 0 0 0 0 0 99676819
2019-04-30 12:14:30 00000000600098D8 test excl. parent l0 0 4 19 13 13 0 0 0 0 0 99676819
--//第3个session 没有阻塞，仅仅gets增加1。
2019-04-30 12:14:31 00000000600098D8 test excl. parent l0 0 4 20 13 13 0 0 0 0 0 99676819
2019-04-30 12:14:32 00000000600098D8 test excl. parent l0 0 4 20 13 13 0 0 0 0 0 99676819
2019-04-30 12:14:33 00000000600098D8 test excl. parent l0 0 4 20 13 13 0 0 0 0 0 99676819
2019-04-30 12:14:34 00000000600098D8 test excl. parent l0 0 4 20 13 13 0 0 0 0 0 99676819
2019-04-30 12:14:35 00000000600098D8 test excl. parent l0 0 4 20 13 13 0 0 0 0 0 99676819
2019-04-30 12:14:36 00000000600098D8 test excl. parent l0 0 4 20 13 13 0 0 0 0 0 99676819
2019-04-30 12:14:37 00000000600098D8 test excl. parent l0 0 4 20 13 13 0 0 0 0 0 99676819
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='test excl. parent l0';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- --------------------- ------ ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-04-30 12:16:02 00000000600098D8 test excl. parent l0 0 4 20 13 13 0 0 0 0 0 99676819
--//最终gets增加3次,MISSES增加1次,SLEEPS增加1次.你可能发现一个现象spin_gets并没有变化。
--//我前面测试已经说明，latch spin方式发生了变化，仅仅spin 10*_spin_count(对于exclusive latch)，_spin_count(对于shared latch).
--//然后在申请不成功的情况下系统调用semop休眠，等待唤醒。这样方式消耗CPU资源很少。
--//那么什么情况下spin_gets下会增加呢?是在spin过程中获得latch吗？如何测试呢？
3.测试spin_gets变化的情况:
--//我以前测试_mutex_spin_count参数时255,65535时，2个semtimedop间隔大约0.001秒。
--//0.0120157-0.0110002 = .0010155
$ cat y2.sh
#! /bin/bash
zdate=$(date '+%Y%m%d%H%M%S')
echo $zdate
source peek.sh 'test excl. parent l0' 6 | timestamp.pl > /tmp/peekx_${zdate}.txt &
seq 20 | xargs -I{} echo -e 'sqlplus -s -l / as sysdba <<< @latch_free\nsleep 1' | bash > /tmp/latch_free_${zdate}.txt &
sqlplus -s -l / as sysdba <<EOF > /tmp/latch_sum_${zdate}.txt &
$(seq 20 | xargs -I {} echo @aa)
EOF
# 参数如下: @ exclusive_latch.txt latch_name willing why where sleep_num
sqlplus / as sysdba <<< "@ exclusive_latch.txt 'test excl. parent l0' 1 1 2 2" > /dev/null &
sleep $1
sqlplus / as sysdba <<< "@ exclusive_latch.txt 'test excl. parent l0' 1 3 4 2" > /dev/null &
wait
--//$1=1.999,不行!!脚本控制太困难使用手工测试看看.
4.手工测试更容易一下。
--//session 1:
SYS@book> @ spid
SID SERIAL# PROCESS SERVER SPID PID P_SERIAL# C50
---------- ---------- ------------------------ --------- ------ ------- ---------- --------------------------------------------------
295 3 43362 DEDICATED 19499 21 2 alter system kill session '295,3' immediate;
SYS@book> @ exclusive_latch.txt 'test excl. parent l0' 1 1 2 100000
LADDR
----------------
00000000600098D8
Statement processed.
Function returned 1
--//session 2:
SYS@book> @ spid
SID SERIAL# PROCESS SERVER SPID PID P_SERIAL# C50
---------- ---------- ------------------------ --------- ------ ------- ---------- --------------------------------------------------
58 21 20098 DEDICATED 20099 28 6 alter system kill session '58,21' immediate;
--//记下spid=20099
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='test excl. parent l0';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-05-05 09:37:53 00000000600098D8 test excl. parent l0 0 4 9 3 3 0 0 0 0 0 46203
--//window 3:
$ gdb -p 20099
(gdb) break kslges
Breakpoint 1 at 0x93f9b74
--//参考：http://blog.itpub.net/267265/viewspace-2641872/ =>[20190418]exclusive latch spin count.txt
(gdb) bt 6
#0 0x00000037990d6407 in semop () from /lib64/libc.so.6
#1 0x0000000009809c0f in sskgpwwait ()
#2 0x00000000098089ce in skgpwwait ()
#3 0x00000000093f9fe1 in kslges ()
#4 0x00000000093f997a in kslgetl ()
#5 0x0000000007d7402e in skdxcall ()
(More stack frames follow...)
--//如果停在kslges这里，进入spin阶段。
(gdb) disassemble kslges
--//反汇编看看.
...
0x00000000093f9ddc <kslges+620>: xor %esi,%esi
0x00000000093f9dde <kslges+622>: mov %esi,-0xd8(%rbp)
0x00000000093f9de4 <kslges+628>: mov %sil,-0xa6(%rbp)
0x00000000093f9deb <kslges+635>: mov 0x0(%r13),%rdi
0x00000000093f9def <kslges+639>: test %rdi,%rdi
0x00000000093f9df2 <kslges+642>: je 0x93fa6c0 <kslges+2896>
0x00000000093f9df8 <kslges+648>: add $0xffffffffffffffff,%ecx
0x00000000093f9dfb <kslges+651>: add $0xffffffffffffffff,%eax
0x00000000093f9dfe <kslges+654>: jne 0x93f9ddc <kslges+620>
...
--//再次说明我不懂汇编,不过可以猜测:
--//%r13对应地址600098D8的内容赋值给%rdi,实际赋值等于session 1的pid(session 1持有该latch)。
--//也就是oradebug peek 0x600098D8 4 看到的内容。
--//如果不等(说明该latch已经释放),执行0x00000000093f9df2 <kslges+642>: je 0x93fa6c0 <kslges+2896>，
--//跳转0x93fa6c0地址执行，跳出spin的循环体。
--//继续：
--//session 2:
SYS@book> @ exclusive_latch.txt 'test excl. parent l0' 1 3 4 2
--//挂起!!
--//window 3:
(gdb) c
Continuing.
Breakpoint 1, 0x00000000093f9b74 in kslges ()
(gdb) ni 1000
0x00000000093f9dfb in kslges ()
(gdb) info register rax rcx rip r13 rdi
rax 0x4dc0 19904
rcx 0x4dbe 19902
rip 0x93f9dfb 0x93f9dfb <kslges+651>
r13 0x600098d8 1610651864
rdi 0x15 21
--//rdi记录的持有latch的pid号.看session 1的@spid输出.
--//session 3:
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='test excl. parent l0';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-05-05 09:41:23 00000000600098D8 test excl. parent l0 0 4 9 3 3 0 0 0 0 0 46203
--//session 1:
--//按ctrl+c中断sleep执行,释放latch.
SYS@book> @ exclusive_latch.txt 'test excl. parent l0' 1 1 2 100000
LADDR
----------------
00000000600098D8
Statement processed.
Function returned 1
Function returned 0
SYS@book> oradebug peek 0x00000000600098D8 8
[0600098D8, 0600098E0) = 00000000 00000000
--//已经为00000000 00000000
--//window 3:
--//单步执行的情况:
(gdb) ni
0x00000000093f9dfe in kslges ()
(gdb) info register rax rcx rip r13 rdi
rax 0x4dbf 19903
rcx 0x4dbe 19902
rip 0x93f9dfe 0x93f9dfe <kslges+654>
r13 0x600098d8 1610651864
rdi 0x15 21
(gdb) ni
0x00000000093f9ddc in kslges ()
(gdb) info register rax rcx rip r13 rdi
rax 0x4dbf 19903
rcx 0x4dbe 19902
rip 0x93f9ddc 0x93f9ddc <kslges+620>
r13 0x600098d8 1610651864
rdi 0x15 21
(gdb) ni
0x00000000093f9dde in kslges ()
(gdb) info register rax rcx rip r13 rdi
rax 0x4dbf 19903
rcx 0x4dbe 19902
rip 0x93f9dde 0x93f9dde <kslges+622>
r13 0x600098d8 1610651864
rdi 0x15 21
(gdb) ni
0x00000000093f9de4 in kslges ()
(gdb) info register rax rcx rip r13 rdi
rax 0x4dbf 19903
rcx 0x4dbe 19902
rip 0x93f9de4 0x93f9de4 <kslges+628>
r13 0x600098d8 1610651864
rdi 0x15 21
(gdb) ni
0x00000000093f9deb in kslges ()
(gdb) info register rax rcx rip r13 rdi
rax 0x4dbf 19903
rcx 0x4dbe 19902
rip 0x93f9deb 0x93f9deb <kslges+635>
r13 0x600098d8 1610651864
rdi 0x15 21
(gdb) ni
0x00000000093f9def in kslges ()
(gdb) info register rax rcx rip r13 rdi
rax 0x4dbf 19903
rcx 0x4dbe 19902
rip 0x93f9def 0x93f9def <kslges+639>
r13 0x600098d8 1610651864
rdi 0x0 0
--//rdi现在是0了.
(gdb) ni
0x00000000093f9df2 in kslges ()
(gdb) info register rax rcx rip r13 rdi
rax 0x4dbf 19903
rcx 0x4dbe 19902
rip 0x93f9df2 0x93f9df2 <kslges+642>
r13 0x600098d8 1610651864
rdi 0x0 0
(gdb) ni
0x00000000093fa6c0 in kslges ()
(gdb) info register rax rcx rip r13 rdi
rax 0x4dbf 19903
rcx 0x4dbe 19902
rip 0x93fa6c0 0x93fa6c0 <kslges+2896>
r13 0x600098d8 1610651864
rdi 0x0 0
(gdb) ni
0x00000000093fa6c2 in kslges ()
(gdb) info register rax rcx rip r13 rdi
rax 0x4dbf 19903
rcx 0x4dbe 19902
rip 0x93fa6c2 0x93fa6c2 <kslges+2898>
r13 0x600098d8 1610651864
rdi 0x0 0
--//现在已经调出循环体.
--//session 3:
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='test excl. parent l0';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-05-05 09:52:20 00000000600098D8 test excl. parent l0 0 4 9 3 3 0 0 0 0 0 46203
--//window 3:
--//不再单步跟踪了.
(gdb) c
Continuing.
--//session 3:
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='test excl. parent l0';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-05-05 09:53:29 00000000600098D8 test excl. parent l0 0 4 10 4 3 0 0 0 0 1 46203
--//注意看现在spin_gets增加了1,在获取latch后,统计信息增加.MISSES增加1,sleeps不变,SPIN_GETS从0->1.
--//另外注意一个细节wait_time时间不变.也就是spin阶段获取latch,等待的时间不计入该视图,仅仅在sleep阶段wait_time才会增加.
--//重复执行前面的y1.sh测试脚本对比就可以发现这个变化:
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='test excl. parent l0';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-05-05 09:58:38 00000000600098D8 test excl. parent l0 0 4 10 4 3 0 0 0 0 1 46203
$ . y1.sh
20190505095849
[4]- Done sqlplus / as sysdba <<< "@ exclusive_latch.txt 'test excl. parent l0' 1 1 2 6" > /dev/null
[5]+ Done sqlplus / as sysdba <<< "@ exclusive_latch.txt 'test excl. parent l0' 1 3 4 6" > /dev/null
[1] Done source peek.sh 'test excl. parent l0' 20 | timestamp.pl > /tmp/peekx_${zdate}.txt
[4]+ Done sqlplus / as sysdba <<< "@ exclusive_latch.txt 'test excl. parent l0' 1 5 6 6" > /dev/null
[2]- Done seq 20 | xargs -I{} echo -e 'sqlplus -s -l / as sysdba <<< @latch_free\nsleep 1' | bash > /tmp/latch_free_${zdate}.txt
[3]+ Done sqlplus -s -l / as sysdba > /tmp/latch_sum_${zdate}.txt <<EOF
$(seq 20 | xargs -I {} echo @aa)
EOF
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='test excl. parent l0';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-05-05 09:59:20 00000000600098D8 test excl. parent l0 0 4 13 5 4 0 0 0 0 1 4050042
--//gets增加3次,misses增加1次.sleeps增加1,spin_gets不变.wait_time,4050042-46203 = 4003839 ,这个单位是微秒.基本吻合.
5.总结:
--//可以通过测试了解latch统计信息的一些细节.
1.获取latch后gets增加.统计信息也在这个时刻更新.
2.正常情况下MISSES=SLEEPS+SPIN_GETS.
3.我在一台生产系统发现misses,sleeps+spin_gets 差别很小,当然如果运行好几年还是有一定差别的(一些中断异常退出会导致这样的差别)
SYS@xxx> select STARTUP_TIME from v$instance ;
STARTUP_TIME
-------------------
2016-08-28 19:15:56
--//已经运行好几年的一个中型数据库的情况.
SELECT SYSDATE
,addr
,name
,level#
,latch#
,gets
,misses
,sleeps
,immediate_gets
,immediate_misses
,waiters_woken
,waits_holding_latch
,spin_gets
,wait_time
,misses-sleeps-spin_gets
FROM v$latch_children
WHERE name LIKE 'cache buffers chains';
..
select count(*) from (
SELECT SYSDATE
,addr
,name
,level#
,latch#
,gets
,misses
,sleeps
,immediate_gets
,immediate_misses
,waiters_woken
,waits_holding_latch
,spin_gets
,wait_time
,misses-sleeps-spin_gets
FROM v$latch_children
WHERE name LIKE 'cache buffers chains'
and misses-sleeps-spin_gets<>0);
COUNT(*)
----------
39
select count(*) from (
SELECT SYSDATE
,addr
,name
,level#
,latch#
,gets
,misses
,sleeps
,immediate_gets
,immediate_misses
,waiters_woken
,waits_holding_latch
,spin_gets
,wait_time
,misses-sleeps-spin_gets
FROM v$latch_children
WHERE name LIKE 'cache buffers chains'
and misses-sleeps-spin_gets=0);
COUNT(*)
----------
8153
--//39 + 8153 = 8192.另外我发现cbc latch 的统计中immediate_gets有数值,说明一些情况可以以nowilling模式获取该拴锁.一些细节还给探究看看.
6.补充测试:
--//测试process allocation latch的情况,我前面说了这个latch很特殊,缺省属于不同类
SYS@book> select CLASS_KSLLT,decode(CLASS_KSLLT,2,KSLLTNAM) name,count(*) from x$kslltr group by CLASS_KSLLT,decode(CLASS_KSLLT,2,KSLLTNAM);
CLASS_KSLLT NAME COUNT(*)
----------- ---------------------------------------- ----------
0 581
2 process allocation 1
SYS@book> select * from x$ksllclass ;
ADDR INDX INST_ID SPIN YIELD WAITTIME SLEEP0 SLEEP1 SLEEP2 SLEEP3 SLEEP4 SLEEP5 SLEEP6 SLEEP7
---------------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------- ----------
00000000861986C0 0 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000
00000000861986EC 1 1 20000 0 1 1000 1000 1000 1000 1000 1000 1000 1000
0000000086198718 2 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000
0000000086198744 3 1 20000 0 1 1000 1000 1000 1000 1000 1000 1000 1000
0000000086198770 4 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000
000000008619879C 5 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000
00000000861987C8 6 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000
00000000861987F4 7 1 20000 0 1 8000 8000 8000 8000 8000 8000 8000 8000
8 rows selected.
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='process allocation';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-05-05 11:05:55 0000000060009F88 process allocation 7 8 82 0 0 37 0 0 0 0 0
$ cat y3.sh
#! /bin/bash
vdate=$(date '+%Y%m%d%H%M%S')
echo $vdate
#source peek.sh "$1" 8 | timestamp.pl >| /tmp/peekx_${vdate}.txt &
#seq 8 | xargs -I{} echo -e 'sqlplus -s -l / as sysdba <<< @latch_free\nsleep 1' | bash >| /tmp/latch_free_${vdate}.txt &
#sleep 1
# 参数如下: @ exclusive_latch.txt latch_name willing why where sleep_num
sqlplus / as sysdba @ exclusive_latch.txt "$1" 1 1 2 3 > /dev/null &
sleep 1
strace -ftT -o /tmp/pp_${vdate}.txt sqlplus -s -l scott/book <<< 'select sysdate from dual ;' &
wait
$ . y3.sh 'process allocation'
20190505110558
[21]+ Stopped sqlplus / as sysdba @ exclusive_latch.txt "$1" 1 1 2 3 > /dev/null
SYSDATE
-------------------
2019-05-05 11:06:01
[22] Done strace -ftT -o /tmp/pp_${vdate}.txt sqlplus -s -l scott/book <<< 'select sysdate from dual ;'
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='process allocation';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-05-05 11:06:04 0000000060009F88 process allocation 7 8 88 1 237 39 0 0 0 0 1962925
--//sleeps=237次.gets什么增加6次.
$ grep "select(0" /tmp/pp_20190505110558.txt |wc
237 2844 17301
--//精确吻合.
$ grep "select(0" /tmp/pp_20190505110558.txt | awk '{print $NF}' | tr -d '<>'| xargs | sed 's/ /+/g' | bc -l
1.930709
--//与后面的等待时间也很接近.
--//我有重复1次.
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='process allocation';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-05-05 11:12:33 0000000060009F88 process allocation 7 8 94 1 237 42 0 0 0 0 1962925
$ . y3.sh 'process allocation'
20190505111240
[22]+ Stopped sqlplus / as sysdba @ exclusive_latch.txt "$1" 1 1 2 3 > /dev/null
SYSDATE
-------------------
2019-05-05 11:12:43
[23] Done strace -ftT -o /tmp/pp_${vdate}.txt sqlplus -s -l scott/book <<< 'select sysdate from dual ;'
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='process allocation';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-05-05 11:12:49 0000000060009F88 process allocation 7 8 98 2 469 44 0 0 0 0 3882339
$ grep "select(0" /tmp/pp_20190505111240.txt |wc
232 2784 16936
--//这次能对上.也许测试有一些干扰.
--//另外我打开一个新会话,视乎gets,IMMEDIATE_GETS都增加,视乎开始以nowilling等待申请.
--//登录前:
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='process allocation';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-05-05 11:17:06 0000000060009F88 process allocation 7 8 115 2 469 50 0 0 0 0 3882339
--//登录后:
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='process allocation';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-05-05 11:17:18 0000000060009F88 process allocation 7 8 116 2 469 51 0 0 0 0 3882339
--//gets,IMMEDIATE_GETS增加1.
--//退出后:
SYS@book> select sysdate,addr,name,level#,latch#,gets,misses,sleeps,immediate_gets,immediate_misses,waiters_woken,waits_holding_latch,spin_gets,wait_time from v$latch_parent where name='process allocation';
SYSDATE ADDR NAME LEVEL# LATCH# GETS MISSES SLEEPS IMMEDIATE_GETS IMMEDIATE_MISSES WAITERS_WOKEN WAITS_HOLDING_LATCH SPIN_GETS WAIT_TIME
------------------- ---------------- ---------------------------------------- ---------- ---------- ---------- ---------- ---------- -------------- ---------------- ------------- ------------------- ---------- ----------
2019-05-05 11:17:26 0000000060009F88 process allocation 7 8 117 2 469 51 0 0 0 0 3882339
--//乱...
[20190505]关于latch 一些统计信息.txt的更多相关文章
- [20190423]简单测试latch nowilling等待模式.txt
[20190423]简单测试latch nowilling等待模式.txt --//我对这个问题的理解就是如果参数willing=0,表示无法获取该latch,直接退出,再寻找类似的latch.--/ ...
- 10G之后统计信息收集后为什么执行计划不会被立马淘汰
在10G之前,使用DBMS_STATS收集统计信息将会导致与此对象相关的游标失效,下次执行此 的时候将会进行HARD PARSE,除非收集的时候NO_INVALIDATE设置为TRUE. 由于硬解析会 ...
- PLSQL_统计信息系列09_统计信息在不同数据库中迁移
2014-01-05 Created By BaoXinjian
- 取得Linux系统的各种统计信息
本文基于Linux 2.6.x内核 一.取得CPU信息(相关文件/proc/stat) 在一个系统中的/proct/stat文件内容如下 $ cat /proc/stat cpu 1039426 17 ...
- SQL Server 统计信息更新时采样百分比对数据预估准确性的影响
为什么要写统计信息 最近看到园子里有人写统计信息,楼主也来凑热闹. 话说经常做数据库的,尤其是做开发的或者优化的,统计信息造成的性能问题应该说是司空见惯. 当然解决办法也并非一成不变,“一招鲜吃遍天” ...
- SQL Server 执行计划利用统计信息对数据行的预估原理二（为什么复合索引列顺序会影响到执行计划对数据行的预估）
本文出处:http://www.cnblogs.com/wy123/p/6008477.html 关于统计信息对数据行数做预估,之前写过对非相关列(单独或者单独的索引列)进行预估时候的算法,参考这里. ...
- SQL Server2016 新功能实时查询统计信息
SQL Server2016 新功能实时查询统计信息 很多时候有这样的场景,开发抱怨DBA没有调优好数据库,DBA抱怨开发写的程序代码差,因此,DBA和开发都成为了死对头,无法真正排查问题. DBA只 ...
- OstrichNet 简易统计信息收集工具
Ostrich 是twitter用于监控服务器性能的一个scala库,项目地址https://github.com/twitter/ostrich, 主要功能是收集.展示统计信息, 同时也提供了关闭服 ...
- DDD实践问题之 - 关于论坛的帖子回复统计信息的更新的思考
之前,在用ENode开发forum案例时,遇到了关于如何实现论坛帖子的回复的统计信息如何更新的问题.后来找到了自己认为比较合理的解决方案,分享给大家.也希望能和大家交流,擦出更多的火花. 论坛核心领域 ...
随机推荐
- semantic功能介绍
semantic功能介绍 gnu Semantic Manual 1,代码自动补全 3,代码导航 启动semantic功能:(semantic-mode 1) 1,Semantic mode 是辅助模 ...
- CodeForces - 1007A （思维+双指针）
题意 https://vjudge.net/problem/CodeForces-1007A 对一个序列重排,使得新的数比原来的数大对应的位置个数最多. 思路 举个栗子,比如1 2 2 3 3 3 3 ...
- 组装数据- 对象里面是key:value, value里面是数组的形式，如 {key:[aa,bb], key:[cc,dd]}
组合后 对象里面是key:value,value里面是数组的形式{key:[aa,bb], key:[cc,dd]} var chinaGeoCoordMap = { '无锡市': [121.4648 ...
- Centos 7 编译安装mariadb 5.5
一.环境 OS :Linux 3.10.0-693.el7.x86_64 mariadb下载地址: ]# wget https://downloads.mariadb.org/interstitial ...
- KPConv针对Modelnet40的分类
1. 训练样本airplane_0001.txt的可视化: 飞机尺度: 物体类别与对应标签: 2. 对训练样本进行降采样: 体素法降采样,降采样的网络大小设置为0.02m.在pycharm下面的Con ...
- ESP8266 AT指令开发(基于STC89C52单片机): 测试下诱人的程序（51单片机，8266，MQTT远程通信控制）
前言 实现的功能,APP通过SmartConfig给Wi-Fi模块配网并绑定设备,然后通过MQTT远程控制开发板的继电器, 简而言之: 51单片机+ESP8266用AT指令实现实现MQTT,(连接的本 ...
- ASP.NET开发实战——（八）ASP.NET MVC 与数据库之MySQL
之前介绍了My Blog如何使用ADO.NET来访问SQL Server获取数据.本章将介绍如何使用My SQL来完成数据管理. 在使用My SQL之前需确保开发环境中安装了My SQL数据库和Con ...
- Asp.Net Core 工作单元 UnitOfWork UOW
Asp.Net Core 工作单元示例 来自 ABP UOW 去除所有无用特性 代码下载 : 去除所有无用特性版本,原生AspNetCore实现 差不多 2278 行代码: 链接:https://pa ...
- 【08月20日】A股滚动市净率PB历史新低排名
2010年01月01日 到 2019年08月20日 之间,滚动市净率历史新低排名. 上市三年以上的公司,2019年08月20日市净率在30以下的公司. 来源:A股滚动市净率(PB)历史新低排名. 1 ...
- Python连载19-装饰器
一.检视一个函数相同的另一种方法 利用属性:函数._name def hello(): print("我是一个测试程序") f = hello print(f.__name__) ...